Make cvxpy optional

README.md

@@ -11,12 +11,20 @@ The library will be uploaded to pypi once the API is stable. Meanwhile, it can b
 
 ### Minimal setup
 
-With the minimal dependencies, corneto can be used with CVXPY and the Coin-OR CBC solver:
+By default, CORNETO does not include any backend nor solver, in order to avoid issues with architectures for which some of the required binaries are not available by default. To install only CORNETO with mininal dependencies, d
 
 ```bash
 conda create --name corneto python=3.8
 conda activate corneto
-pip install corneto-0.9.0a0-py3-none-any.whl
+pip install corneto-0.9.0a3-py3-none-any.whl
+```
+
+#### CVXPY backend
+
+CVXPY supports many solvers, including the Coin-OR CBC solver through the `cylp` package. If you want to have support for solving problems both for non-commercial and commercial projects, you can use CORNETO with `cvxpy` and `cylp`:
+
+```bash
+pip install cvxpy cylp
 ```
 
 To test if the solver is correctly installed, test the following command on the new environment:
@@ -26,16 +34,25 @@ conda activate corneto
 python -c "import cvxpy; print(cvxpy.installed_solvers())"
 ```
 
-The `CBC` solver should appear on the list.
+The CBC solver should appear on the list. Please see the CVXPY documentation for more information on how to install other solvers https://www.cvxpy.org/install/. Depending on the solver and the platform, you will need to take different actions. For example, for using Gurobi, you will need to install the Gurobi solver with a valid license and then installing `gurobipy` dependency on the environment.
+
+#### PICOS backend
+If you want to use `PICOS` instead of `CVXPY` backend:
+
+```bash
+pip install PICOS
+```
+
+Check `PICOS` requirements for different solvers here: https://picos-api.gitlab.io/picos/introduction.html#features
 
 ### Complete setup
 
 A recommended setup for testing and generating plots with corneto requires additional dependencies. A recommended environment is created with the following command:
 
 ```bash
-conda create --name corneto python=3.8 matplotlib=3.5.1 networkx=2.7.1 pandas=1.4.3 jupyter=1.0.0 pydot=1.4.1 graphviz=2.50.0
+conda create --name corneto python=3.8 cvxpy=1.2.1 cylp=0.91.5 matplotlib=3.5.1 networkx=2.7.1 pandas=1.4.3 jupyter=1.0.0 pydot=1.4.1 graphviz=2.50.0
 conda activate corneto
-pip install corneto-0.9.0a0-py3-none-any.whl
+pip install corneto-0.9.0a3-py3-none-any.whl
 ```
 
 Now the notebook included in the `tests` folder should be able to be run:
@@ -44,7 +61,3 @@ Now the notebook included in the `tests` folder should be able to be run:
 cd corneto
 jupyter nbconvert --execute --to html tests/notebooks/tutorial.ipynb
 ```
-
-### Aditional solvers
-
-Please see the CVXPY documentation for more information on how to install other solvers https://www.cvxpy.org/install/. Depending on the solver and the platform, you will need to take different actions. For example, for using Gurobi, you will need to install the Gurobi solver with a valid license and then installing `gurobipy` dependency on the environment.

corneto/__init__.py

@@ -1,4 +1,4 @@
 from corneto._constants import *
 from corneto.core import ReNet
 
-__version__ = '0.9.0-alpha.2'
+__version__ = '0.9.0-alpha.3'

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "corneto"
-version = "0.9.0-alpha.2"
+version = "0.9.0-alpha.3"
 description = "CORNETO: CORe NETwork Optimizer, a MILP optimization library for Biochemical Networks"
 authors = ["Pablo Rordriguez-Mier <[email protected]>"]
 classifiers = [
@@ -19,7 +19,7 @@ include = ["LICENSE"]
 [tool.poetry.dependencies]
 python = ">=3.8 <3.11"
 numpy = "^1.22.3"
-cvxpy = "^1.2.1"
+cvxpy = {version = "^1.2.1", optional = true}
 cylp = {version = "^0.91.5", optional = true}
 networkx = {version = "~2.7.1", optional = true}
 PICOS = {version = "^2.4.1", optional = true}
@@ -37,8 +37,12 @@ mkdocs-material = "^7.2.1"
 mkdocstrings = "^0.15.2"
 
 [tool.poetry.extras]
-cbc = ["cylp"]
-extras = ["PICOS", "matplotlib", "networkx"]
+cvxpy = ["cvxpy"]
+picos = ["PICOS"]
+all_backends = ["cvxpy", "PICOS"]
+gurobi = ["gurobipy", "PICOS"]
+cbc = ["cvxpy", "cylp"]
+full = ["cvxpy", "PICOS", "matplotlib", "networkx"]
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]